I agree that 10-15 references is totally bogus; I said so much in a reply to a different comment. I'm still boggling though, actually more so than I was before

I agree that 10-15 references is totally bogus; I said so much in a reply to a different comment. I'm still boggling though, actually more so than I was before your reply. You mention examples here that make the whole process seem incredibly error-prone, bound to result in false positives and negatives, and as you described it fraught with peril. This is a discouraging testament to the utility of doing reference checks. I feel quite uneasy questioning years of standard hiring practice, but do the benefits of this sufficiently outweigh the drawbacks? You gather data about the candidate, but no metadata about the recommendation, which I can see make a considerable difference in how the reference is acted on. Metadata such as: * the relationship dynamics of the recommender to other people besides the candidate * how candid the recommender is about the benefits and drawbacks of things he/she evaluates * what sort of events the recommender encountered, and what mood that set him/her in before talking with you. All of these would put the recommendation in better context. There are probably other data points (recommendations of the recommenders, etc), that could be considered, and then there are time/benefit trade offs to decide which are best left unconsidered. I will agree that a reference check will likely shuffle out someone who is obviously a raving lunatic, and would never work out no matter where they are hired. But, beyond that, it seems more nebulous. I have the sense that there is a combination of institutional inertia and real benefit that drives this practice; I want to understand where the bounds of each are, and whether my intuition is completely off-base.